Microsoft 365 is back. Copilot waited on Exchange
Microsoft's record for MO1550821 starts November 12, 2026. A network-capacity upgrade, not a cyberattack, took Exchange-backed Outlook and Copilot down.

Microsoft's record for MO1550821 starts November 12, 2026. A network-capacity upgrade, not a cyberattack, took Exchange-backed Outlook and Copilot down.

Microsoft tracked the outage as MO1550821. The University of Pennsylvania's copy of Microsoft's service-health messages ↗ lists a start of Thursday, November 12, 2026, at 10:05 UTC (5:05 a.m. ET) and an end of Friday, November 13, at 4:10 p.m. UTC (11:10 a.m. ET). Status on that page: service restored.
The picture above is ours.
CNBC ↗, updated Saturday, November 14, says engineers were working on connectivity problems that began shortly before 1 p.m. ET on Thursday. That is a later clock than 10:05 UTC. Both are in print. Microsoft's incident record is the earlier one. CNBC is describing when the disruption was being worked.
The clocks stay split at the end, too. The service-health end time is Friday at 4:10 p.m. UTC. A final-status note on that same record, posted Saturday at 4:17 a.m. UTC, still said Microsoft was restoring service to remaining customers and regions. Reuters ↗, datelined Saturday, says Microsoft then posted that it had "completed our efforts to address the underlying issue of the incident and restored service to the remaining impacted customers and regions." The Guardian ↗ describes a shorter public stretch: roughly 18 hours before recovery started, just before midday GMT on Friday.
The Friday, November 13, 5:57 p.m. UTC service-health update says: "We are deeply sorry for the disruption this incident has caused you and your business." The same update says: "This was a technical issue not related to cyber or AI / agents. The incident was introduced by an issue during an earlier network capacity upgrade that surfaced following a recent network change."
CNBC says Thursday's updates blamed servers that failed automated health checks, with the remaining machines overloaded. The Saturday final status promises a preliminary post-incident report within two business days and a final report within five. This piece does not include that report.
Charles Lamanna, whom the Guardian identifies as Microsoft's executive vice-president of Copilot, told the Guardian: "We regret that today's outages caused disruptions for our customers, and we are working to address the issue."
Microsoft's final status says features that talk to Exchange Online were hit, and features that do not talk to Exchange stayed up.
| Service | Microsoft's record |
|---|---|
| Outlook | Web, desktop, ActiveSync, and REST access failed, or mail was delayed |
| Copilot | Chat load, and access to mail, calendar, and files |
| Teams core | Calls, meetings, group chats, and standard channels stayed up |
| Teams extras | Calendar edits, live events, and Copilot answers over mail failed |
| SharePoint | Files stored there stayed up. Some external sharing failed |
Graph calls that read Exchange mail, calendar, or contacts were on the impact list, and so were some Teams membership calls and Defender XDR portal pages. Moneycontrol ↗ names Exchange Online plus dependent features in Teams, Microsoft 365 Copilot, SharePoint Online, and Defender XDR, under the same incident id.
The Guardian says OpenAI was investigating broken Microsoft connectors, with new connections failing and tools misbehaving, and that Anthropic had identified an upstream-provider issue affecting its Microsoft 365 connector. A drafting box with no live mailbox was in a different position from an assistant whose tools read Exchange.
Moneycontrol says Downdetector logged nearly 130,000 Outlook reports at the peak and more than 47,000 for Microsoft 365. The Guardian says Down Detector showed nearly 130,000 reports of problems in 74 countries. Reuters, citing Ookla, says Downdetector logged more than 7.4 million reports worldwide and peaked near 20,000 a minute. Those are three published counts. We did not count the reports.
Reuters says the outage spilled into Asia-Pacific business hours, and that organizations hosting Microsoft services on their own servers were somewhat spared. The Guardian says British police forces told people to phone instead of reporting online, and that some UK schools told parents to phone about attendance.
"Keep a backup comms channel ready." — GoatKing ↗ (@aru2u4), November 15
"Outlook, Teams and Copilot disruptions exposed how dependent the world has become on Microsoft's services." — Theepan ↗ (@Theepan15789)
"An outage is an operations problem." — Xiaoyu Wang ↗ (@wangxiaoyu_X). His post also gives a share-price move. Reuters says Microsoft shares closed flat on Friday, at $523.88. Use the operations line.
"Microsoft 365 became Microsoft 364!" — ScrollTheory ↗ (@TheScrollTheory)
Reddit was not reachable for this story.
Copilot was down where it reads Exchange. Microsoft's own split is the useful one. Teams calls, meetings, group chats, and standard channel messages stayed up. Copilot Chat, mailbox access, and any Copilot answer that needed mail, calendar, or Microsoft 365 files did not. An agent that drafts from a prompt you paste is a different failure mode from an agent whose tools call Graph.
Keep the fallback that does not need that mailbox. GoatKing's line matches what the Guardian reported from police forces and schools: a phone number people already know. Inside Microsoft's own list, Teams chat was the path that stayed up, and calendar invites were not. For a meeting note-taker that does not sit on Exchange, Claude Meetings is a Google Meet beta. Grok Bot can join a Teams call, which is the part Microsoft said stayed up, and it still is not a copy of the mailbox. We did not test either product during this incident.
Use another model only on text you already have. ChatGPT, Claude, and Gemini can draft from a thread you paste. The Guardian's connector note says the live Microsoft 365 hook on the OpenAI and Anthropic sides was part of the breakage. Microsoft 365 Copilot is also not the coding product in our Copilot vs Cursor comparison. Admins should follow MO1550821 in Microsoft 365 service health. The promised post-incident report is the document that will say what the network-capacity upgrade actually changed.